File size: 5,117 Bytes
c5ded30
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
65e8cd5
c5ded30
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2b8b080
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
import gradio as gr
import os
from io import BytesIO
from PIL import Image, ImageDraw, ImageFont
from PIL import ImageColor
import json
import google.generativeai as genai
from google.generativeai import types
from dotenv import load_dotenv


# 1. SETUP API KEY
# ----------------
load_dotenv()
api_key = os.getenv("Gemini_API_Key") 
# Configure the Google AI library
genai.configure(api_key=api_key)


# 2. DEFINE MODEL AND INSTRUCTIONS

bounding_box_system_instructions = """
    Return bounding boxes as a JSON array with labels. Never return masks or code fencing. Limit to 25 objects.
    If an object is present multiple times, name them according to their unique characteristic (colors, size, position, unique characteristics, etc..).
      """
model = genai.GenerativeModel( model_name='gemini-2.5-flash', system_instruction=bounding_box_system_instructions , safety_settings=[ types.SafetySettingDict( category="HARM_CATEGORY_DANGEROUS_CONTENT", threshold="BLOCK_ONLY_HIGH", ) ],)
generation_config = genai.types.GenerationConfig(
        temperature=0.5,
 
    )   


def generate_bounding_boxes(prompt, image):
    image = image.resize((1024, int(1024 * image.height / image.width)))
    response = model.generate_content([prompt, image], generation_config=generation_config)
    bounding_boxes = parse_json(response.text)
    img=plot_bounding_boxes(image, bounding_boxes)
    return img


def parse_json(json_output): 
    lines = json_output.splitlines()
    for i, line in enumerate(lines):
        if line == "```json":
            json_output = "\n".join(lines[i+1:])  # Remove everything before "```json"
            json_output = json_output.split("```")[0]  # Remove everything after the closing "```"
            break
    return json_output

def plot_bounding_boxes(im, bounding_boxes):
    """
    Plots bounding boxes on an image with labels.
    """
    additional_colors = [colorname for (colorname, colorcode) in ImageColor.colormap.items()]

    im = im.copy()
    width, height = im.size
    draw = ImageDraw.Draw(im)
    colors = [
        'red', 'green', 'blue', 'yellow', 'orange', 'pink', 'purple', 'cyan',
        'lime', 'magenta', 'violet', 'gold', 'silver'
    ] + additional_colors

    try:
        # Use a default font if NotoSansCJK is not available
        try:
            font = ImageFont.load_default()
        except OSError:
            print("NotoSansCJK-Regular.ttc not found. Using default font.")
            font = ImageFont.load_default()

        bounding_boxes_json = json.loads(bounding_boxes)
        for i, bounding_box in enumerate(bounding_boxes_json):
            color = colors[i % len(colors)]
            abs_y1 = int(bounding_box["box_2d"][0] / 1000 * height)
            abs_x1 = int(bounding_box["box_2d"][1] / 1000 * width)
            abs_y2 = int(bounding_box["box_2d"][2] / 1000 * height)
            abs_x2 = int(bounding_box["box_2d"][3] / 1000 * width)

            if abs_x1 > abs_x2:
                abs_x1, abs_x2 = abs_x2, abs_x1

            if abs_y1 > abs_y2:
                abs_y1, abs_y2 = abs_y2, abs_y1

            # Draw bounding box and label
            draw.rectangle(((abs_x1, abs_y1), (abs_x2, abs_y2)), outline=color, width=4)
            if "label" in bounding_box:
                draw.text((abs_x1 + 8, abs_y1 + 6), bounding_box["label"], fill=color, font=font)
    except Exception as e:
        print(f"Error drawing bounding boxes: {e}")

    return im
def gradio_interface():
    """
    Gradio app interface for bounding box generation with example pairs.
    """
    # Example image + prompt pairs
    examples = [
        ["cookies.jpg", "Detect the cookies and label their types."],
        ["messed_room.jpg", "Find the unorganized item and suggest action in label in the image to fix them."],
        ["yoga.jpg", "Show the different yoga poses and name them."],
        ["zoom_face.png", "Label the tired faces in the image."]
    ]

    with gr.Blocks(gr.themes.Glass(secondary_hue= "rose")) as demo:
        gr.Markdown("# Gemini Bounding Box Generator")

        with gr.Row():
            with gr.Column():
                gr.Markdown("### Input Section")
                input_image = gr.Image(type="pil", label="Input Image")
                input_prompt = gr.Textbox(lines=2, label="Input Prompt", placeholder="Describe what to detect.")
                submit_btn = gr.Button("Generate")

            with gr.Column():
                gr.Markdown("### Output Section")
                output_image = gr.Image(type="pil", label="Output Image")
                #output_json = gr.Textbox(label="Bounding Boxes JSON")

        gr.Markdown("### Examples")
        gr.Examples(
            examples=examples,
            inputs=[input_image, input_prompt],
            label="Example Images with Prompts"
        )

        # Event to generate bounding boxes
        submit_btn.click(
            generate_bounding_boxes,
            inputs=[input_prompt, input_image],
            outputs=[output_image]
        )

    return demo



if __name__ == "__main__":
    app = gradio_interface()
    app.launch()